I personally have never used what I would call "proper Litz" wire - individually insulated strands woven so they are evenly distributed in location throughout the cable bundle. I have always used roped enamel wire as it is way cheaper and seems to give most of the benefit. Any comments?

Can't say that I have ever really used Litz, but if you need some, you might be interested in Furukawa's TEX ELZ wire, which is Litz wire surrounded by three insulation layers (good for offline stuff).

PACE Components has it, and you can buy it in large or small quantities from their online store. I live in the US, and I think it took about 10 - 11 days for my order to arrive from the UK. The price range is from 0.66 Euros/m to 1.52 Euros/m based on size, which is something like (0.88 USD/m to 2 USD/m). Of course, if you order alot of wire, the price per meter is lower.

I don't know what is the range of frequencies you are working but I can say that HF together with high current require for sure the use of Litz wire, otherwise AC losses in the wire become relevant and can drastically affect the behavior of your SMPS.

Our company used Litze wire in bundles platted by us for a contactless shaft rotating 100W power transformer. Due to distorted flux paths optimising the wire arrangement was critical.

I've done several transformer designs with litz wire. 750 VA to 2000VA push-pull or full bridge, 70kHz to 120kHz, toroids and PQ cores where pin terminations were already impractical due to current. Litz does cost more, but the bigger issue is leadtime. There are some sources out there that stock some litz allowing for acceptable substitution in the prototype stage until the proper-selected litz arrives. Another nice thing is that the winding of litz on the core is very perdictable. Once you've got the transformer down, you can measure, precut, and solder/terminate the litz with the terminals or lugs you're using and THEN manually wind the transformer knowing the terminals will fall exactly where you need them. Also soft and easy on the hands.

However, my experience has all been with the less fancy Type 1 or 2, single or double served litz, and not with the very exotic Type 3 through 8, or custom.

At a previous company we used litz wire for the 20kW resonant converters. Soldering the wire so that all the strands were connected was difficult to get right in production. A single turn of Litz braid was used for the low voltage high current transformers. Very expensive but good performance. When I designed the new high current power supplies I used a phase-shift converter and changed the transformers to ECW rope on the primary and a single turn of foil for the secondary. The foil secondary was almost as good performance as the litz braid, but much lower cost and easier to cool.
My advice - only use litz where high efficiency is needed above all else. Losses can be controlled in other ways: arrangement of windings, core type, waveform shape, position of gaps if it is an inductor e.t.c.

Thanks everybody for your comments, interesting stuff. I wonder if I was less than clear with my question for which I apologies. I don't consider roped Enamelled Copper Wire to be "proper Litz". It is a cheaper and easier to get hold of substitute which as far as I am aware gives most (if not all!) of the advantages in terms of reduced loss at HF for many applications. Foil is of course also used at high current and frequency and gives excellent fill. What I am particularly interested in is to find out whether people have had to resort to "proper Litz" in some applications and if so why. Is the packing of the wire any better? Is the self capacitance any better due to additional insulation between the strands? Is the woven nature better at averaging the placement of the individual strands and therefore gives a further reduction in HF losses?